The invention relates to auto focus methods, and more particularly, to methods for shortening focus time using image evaluation values.
Conventionally, auto focus methods implement processes according to generated evaluation values corresponding to desired image signals to control focus of a lens. This requires shifting a lens module repeatedly back and forth for coarse overall scanning and fine scanning, thus obtaining evaluation values corresponding to each focus locations. Next, according to focus modes, analyses and comparisons are performed to locate an optimum focus location, thereby shifting the lens module to the optimum focus location. Driving a motor, repeatedly, to carry the lens module wastes time and increase wear on the motor and transmission components thereof. If a resulting image is not the desired image, repeat focus processes waste even more time.
The described evaluation values are high frequency components (HFC), generated by images passing a band-pass filter (BPF) or high-pass filter (HPF). An evaluation curve generates a relatively maximal (or minimal) evaluation value when a zoom lens shifts, such that therefore a focus location corresponding to the evaluation value is the focal point. Due to unsatisfactory slope values of the evaluation curve, it is difficult for conventional auto focus methods to obtain the relatively maximal (or minimal) evaluation value due to the external factors, such that the optimum focus location cannot be located. In actual focus processes for a zoom lens, different zoom ratios result in discrepancies in evaluation distribution.
When a zoom lens shifts to a telephotographic or pantoscopic mode, problems occur if the size of a viewscreen corresponding to a focus region is fixed, such as generation of a large peak value with a steep curve when the zoom lens shifts to a telephotographic location, as shown in FIGS. 1A and 1B. Accordingly, a small peak value with a smooth curve may be generated when the zoom lens shifts to a pantoscopic location, as shown in FIGS. 1C and 1D.
One solution is design of a special filter for a specific hardware device, as disclosed in Taiwan patent No. 172155. An auto focus device of the disclosure maintains a slope value of an evaluation curve based on a fixed size of a viewscreen using high frequency components obtained through multiple high frequency filters, thus locating a focal point of the auto focus device.
Additionally, two focus control circuits can be utilized. As disclosed in U.S. Pat. No. 4,903,134, in an automatic focus circuit of a video camera, automatic focus is performed by a focus evaluation value generated in response to a video signal obtained by an image sensing circuit. A first focus motor control circuit controls the rotation of a focus motor so that a focus lens is fixed once in the position where the focus evaluation value takes the maximal value. A second focus motor control circuit changes the focus motor, by a minimum predetermined amount, to determine the slope of the focus evaluation value, which change is repeated until inversion of the slope, that is, the maximal point is detected. As a result, the position of the focus lens is corrected at the maximal point of the focus evaluation value. When the correction amount exceeds a predetermined value, first automatic focus by the first focus motor control circuit is resumed.
Further, as disclosed in U.S. Pat. No. 5,235,428, in an auto focus system, a focus detection signal is generated from a video signal produced by a video camera by deriving higher frequency components of the video signal, detecting when the higher frequency components exceed a minimal threshold, sensing when the video signal exceeds a maximal threshold, and integrating the higher frequency components which exceed the minimal threshold, except during an interval when the video signal exceeds the maximal threshold, thereby producing the focus detection signal.
Additionally, FIG. 2 is a schematic diagram of a curve generated using a mount-climbing focus method. The mount-climbing focus method obtains a peak value of a curve using an asymptotic approximation method. The curve similar to the curves shown in FIGS. 1B and 1D is generated according to HFC accumulation values (i.e. evaluation values) by different focal lengths. A peak value corresponding to a maximal evaluation value on the curve corresponds to a true focal point. Defined images for a focus process depend on the highest point on the curve to be located. If the steep or smooth state described occurs, location of the peak value takes longer and even the peak value may never be ascertained.
Details of the mount-climbing focus method are further described in the following. Referring to FIG. 2, the X axis indicates focus locations (FL) and the Y axis indicates high frequency components of image signals. A start focus location is point A, and a focus motor is driven to point B (if the current direction of the motor is forward). If the high frequency component at point B exceeds the component at point A, the motor is driven to point C in the forward direction. If the high frequency component at the point C exceeds the component at point B, the motor is further driven to point D in the forward direction. As shown in FIG. 2, the high frequency component at point D is smaller than the component at point C, indicating the image captured at point C is clearer than the image captured at point D, such that the motor is reversely driven to point F. If the high frequency component at point E exceeds the component at point C, the motor is further reversely driven to point F. If the high frequency component at point F is smaller than the component at point E, the motor is further driven to point G in the forward direction. The motor is repeatedly driven back and forth to approximate the high frequency components to locate the relatively highest point on the curve, thus obtaining the maximal high frequency component. During the approximation process, if the curve shown in FIG. 2 is steep or smooth as the curve shown in FIG. 1B or 1D, and only if the motor is driven by a smaller unit distance, the maximal high frequency component is difficult to locate, thus increasing the focus time.
Additionally, another auto focus method is a seven-step method, driving a motor by seven steps as unit distances and obtaining high frequency components corresponding thereto. When all the high frequency components are obtained, a single-step method is implemented to one unit distance comprising the maximal high frequency component, thus obtaining a real maximal high frequency component. A drawback of the method is implementation of search operations to a length of a focus track only if search operations implemented to the entire focus track have been completed, using much time.
To compare the seven-step method with other auto focus methods using hardware structures, the mount-climbing focus method is simple and can rapidly locate a focal point, however, must repeatedly drive the motor back and forth to obtain the focal point, thus requiring much time. Thus, an improved auto focus method is desirable for shortened focus time.